ReferenceFull Library ReferenceNemo AutomodelNemo AutomodelSharednemo_automodel.shared.tokenizer_serialization

nemo_automodel.shared.tokenizer_serialization

View as Markdown

Serialization cleanup shared by portable tokenizer exporters.

Module Contents

Functions

NameDescription
_write_jsonWrite tokenizer metadata as formatted JSON.
restore_source_tokenizer_serialization_stateRemove training-time tokenizer state while retaining tokenizer edits.

API

nemo_automodel.shared.tokenizer_serialization._write_json(
path: str,
contents: dict
) -> None

Write tokenizer metadata as formatted JSON.

nemo_automodel.shared.tokenizer_serialization.restore_source_tokenizer_serialization_state(
original_model_path: str | None,
hf_metadata_dir: str,
tokenizer: transformers.PreTrainedTokenizerBase
) -> None

Remove training-time tokenizer state while retaining tokenizer edits.

Parameters:

original_model_path
str | None

Local source snapshot, or None to clear transient state.

hf_metadata_dir
str

Export directory containing the serialized tokenizer.

tokenizer
PreTrainedTokenizerBase

Tokenizer supplying the current padding token.